범용 병렬화일 시스템 상에서 MPI-IO 방안의 성능 평가 벤티마크

Benchmarks for Performance Testing of MPI-IO on the General Parallel File System

  • 발행 : 2001.06.01

초록

IBM developed the MPI-IO, we call it MPI-2, on the General Parallel File System. We designed and implemented various Matrix Multiplication Benchmarks to evaluate its performances. The MPI-IO on the General Parallel File System shows four kinds of data access methods : the non-collective and blocking, the collective and blocking, the non-collective and non-blocking, and the split collective operation. In this paper, we propose benchmarks to measure the IO time and the computation time for the data access methods. We describe not only its implementation but also the performance evaluation results.

키워드

참고문헌

  1. 'An Introduction to GPES 1.2' IBM T. J. Watson Research Center, Dec, 1998. (Source : http://www.rs6000.ibm.com/resource/technology/paper1.html)
  2. 'General Parallel File System: High Performance Parallel File Systems,' IBM T. J. Watson Research Center, Dec, 1998. (Source : http://www.rs6000.ibm.com/resource/software/sp_products/gpfs_perf.html)
  3. 'GPES Performance,' IBM T. J. Watson Research Center, Feb, 1999, (Source : http://www.rs6000.ibm.com/resource/technology/gpfs_perf.html)
  4. IBM Internal Documents, 'Supported Functions of MPI-I/O,' IBM T. J. Watson Research Center, 1998
  5. Message Passing Interface Forum, 'MPI-2 : Extensions to the Message Passing Interface, Univ. of Tennssee,' Jul, 1997
  6. P. S. Pachew, 'Users Guide to MPI,' Univ. of San Francisco, Mar, 1998
  7. R. Bordawekar, J. M. del, Rossario, and A. Choudhary, 'Design and Evaluation of Primitives for Parallel I/O,' Proceedings of Supercomputing 93, pp.452-461, 1993 https://doi.org/10.1109/SUPERC.1993.1263493
  8. R. Thakur, W. Gropp and E. Lusk, 'Achieving High Performance with MPI-IO', ANL/MCS pp.742-0299, Argonne National Laboratory, Feb, 1999